Identification of Hydroxysteroid (17β) dehydrogenase type 12 (HSD17B12) as a CD8⁺ T-cell-defined human tumor antigen of human carcinomas

Human carcinoma and normal cell lines

The HLA-A2⁺ PCI-13 and PCI-30, and HLA-A2^neg PCI-4A SCCHN cell lines used in this study were established at UPCI by Dr. Teresa Whiteside [17]. The HLA-A2⁺ UD-SCC 6 SCCHN cell line was obtained from Dr. H. Bier, Technical University of Munich, Munich, Germany. The HLA-A2⁺ SCCHN SCC-4, and MDA-MB-453, MDA-MB-231, and MCF-7 breast carcinoma cell lines, as well as the HLA-A2⁺ normal human fibroblast cell line, MCR-5, were obtained from American Type Culture Collection (Rockville, MD). The HLA-A2⁺AD10, OVCAR429, and OVCAR3 ovarian carcinoma cell lines were obtained from Dr. S. Khlief (Naval Hospital, Bethesda MD). The JL-BRL 6 cell line is a low-passage human breast tissue cell line established by Dr. Latimer from a reduction mammoplasty [18]. All the cell lines, with the exception of JL-BRL 6, were maintained in a complete medium (CM) consisting of RPMI-1640 medium supplemented with 10% (v/v) FBS, 2 mM l-glutamine, 50 μg/ml streptomycin, and 50 IU/ml penicillin (Life Technologies, Inc., Grand Island, NY). The JL-BRL 6 cell line was maintained in a defined culture medium developed and supplied by Dr. Latimer [18].

Antibodies and peptides

Immunoaffinity-purified mAb used in these studies were obtained from supernatants of hybridomas producing HLA-A,-B,-C antigen (Ag)-specific mAb (clone W6/32, IgG_2a), HLA-DR-specific mAb (clone L243, IgG_2a), and the HLA-A2,-A28 Ag-specific mAb (KS1, IgG₁). The W6/32 and L243 hybridomas were obtained from ATCC. The KS1 and an IgG1 isotype mAb were from Dr. Soldano Ferrone (University of Pittsburgh, Pittsburgh PA) [19]. The peptide-immunoaffinity–purified polyclonal rabbit anti-TYDKIKTGL peptide antibody, which cross-reacts with the IYDKIKTGL (Hsd17b12/HSD17B12_114–122) peptide, has previously been described [16]. Peptides were synthesized using standard Fmoc chemistry and their sequences confirmed by tandem MS.

In vitro stimulation (IVS) of HLA-A2-restricted, anti-HSD17B12_114–122 peptide human CD8⁺ T cells

HSD17B12_114–122 peptide CD8⁺ T-cell effectors were generated following in vitro stimulation (IVS) of CD8⁺ T cells obtained from PBMC of HLA-A2⁺ healthy donors using peptide-pulsed autologous DC as antigen presenting cells (APC) as previously described [20].

INFγ ELISPOT assay

INFγ ELISPOT assays were performed in 96-well flat-bottomed nitrocellulose plates (MAHAS4510; Millipore, Bedford, MA) using, respectively, anti-IFNγ capture mAb and biotinylated anti-IFNγ detection mAb (Mabtech, Inc. Cincinnati, OH), as previously described [20]. The assays consisted of 1 × 10⁴ target cells and CD8⁺ T cells at an E-/T-cell ratio of 2:1. Blocking experiments were performed by incubating target cells with blocking and isotype control mAb (10 μg/ml) for 30 min at 4°C prior to addition of effectors.

Transfection of MCR-5 cells with HSD17B12 cDNA

MRC-5 cells were transfected with HSD17B12 cDNA (SC114479, OriGene Technologies, Inc. Rockville, MD) by electroporation using a Nucleofector device according to the manufacturer’s protocol. The nucleofection efficacy of empty pCMV6-XL5 vector and pCMV6-XL5-HSD17B12–transfected MCR-5 cells was monitored by quantitative reverse transcription-PCR (qRT-PCR) of HSD17B12 mRNA using the following primers designed in this laboratory; forward primer: TTGCTGTTGACTTTGCATCAG; reverse primer: TTCACTAAGATGCCGA TTTCAA and probe: 5′-/56 FAM/TGATAAAATTAAAACAGGCTTGGCTGGT/3BHQ-1/3′.

Immunoblot analyses of HSD17B12 expression in human normal and tumor cell lines

The expression of HSD17B12 in human normal and tumor cell lines was analyzed by immunoblot using the purified rabbit antibody at a concentration of 1 μg/ml and developed using horseradish peroxidase-conjugated goat anti-rabbit IgG Fc fragment-specific antibody (Jackson ImmunoResearch Laboratories, Inc. West Grove, PA) at 1:10,000 dilution and Western Lightning Plus-ECL (Perkin Elmer, Inc., Waltham MA) [16].

Immunohistochemical analysis of human tumor cell lines and normal cells and tissues for HSD17B12 expression

The optimal staining dilution of the peptide immunoaffinity polyclonal rabbit anti-TYDKIKTGL antibody (1 μg/ml) for HSD17B12 was determined by immunofluorescence microscopy using formalin-fixed PCI-13 cells by Dr. Dhir (Director, Department of Pathology, UPMC Shadyside Hospital, Pittsburgh PA) using an Olympus BX-41 microscope. Controls included the use of the blocking peptide. Two formalin-fixed paraffin-embedded tissue microarrays (TMA) also were analyzed for HSD17B12 expression with the peptide-immunoaffinity–purified, polyclonal rabbit anti-TYDKIKTGL antibody using standard procedures. The stained sections were analyzed using an Olympus BX-41 microscope. One TMA consisting of 15 oral cavity SCCHN specimens, which included surrounding mucosa, was constructed in the laboratory of Dr. Dhir from IRB approved excess sections of paraffin blocks of specimens that were initially generated for clinical evaluation. The second TMA, the commercially available SCCHN TMA (cat # HN803, US Biomax Inc, Rockville, MD) was analyzed using a Nikon Eclipse microscope. All stained sections were analyzed and scored by two pathologists to avoid bias, and the average of their scores recorded. The sections were scored according to the % of cells staining (<25%: negative; 25–75%: heterogenous; and >75%: positive), staining intensity (weak, moderate, and strong) and cellular localization (nucleus or cytoplasm).

Small interfering RNA (siRNA) inhibition of HSD17B12 expression

HSD17B12 siRNA (sc-96987) and two control siRNA (sc-37007 and sc-36869) purchased from Santa Cruz Biotechnology, Inc. Santa Cruz, CA) were used to demonstrate siRNA inhibition of the expression of HSD17B12 in PCI-13 cells in a protocol suggested by the manufacturer. To accomplish maximum inhibition of HSD17B12 mRNA synthesis in PCI-13 cells, 5 × 10⁴ cells/well/6-well plates were transfected with 8 μg/siRNA in complete medium without antibiotics. After a 24 h incubation, supernatant was removed and serum-free RPMI-1640 medium added with or without 1 μM arachidonic acid (MP Biomedicals, Solon, OH) (1:10 dilution of a 1:100 dilution in PBS of a 1 mM AA/DMSO stock solution) or 1nM estradiol (Sigma, St. Louis) (1:10 dilution of a 1:100 dilution in PBS of a 1 μM E2/ethanol stock solution). After 48 h incubation, cells were harvested for analysis. Fluorescein isothiocyanate-conjugated control siRNA-A (sc-36869) and control siRNA-A (sc-37007) were the controls. The knockdown of HSD17B12 expression in PCI-13 was monitored by qRT-PCR relative to reporter gene β-glucuronidase (βGUS) mRNA using primers listed above for HSD17B12 and those previously described [20]. The effect of HSD17B12 knockdown on growth of transfected cells PCI-13 cells cultured in the absence or presence of 1 μM AA or 1nM E2 for 48 h was determined by harvesting and determining the number of viable cells (cells excluding trypan blue).

Statistical analysis

The results of ELISPOT assays and siRNA experiments were analyzed using Students’ t-tests and considered significant at P < 0.05. The significance of the results of staining the TMA relative to clinicopathological characteristics of the specimens was determined using Pearson Correlation Asymp Sig. (2-sided) analysis.

Generation of HSD17B12 peptide-reactive CD8⁺ T cells

The HSD17B12_114–122 peptide was tested for its ability to induce the expansion of peptide-specific CD8⁺ T cells following IVS of CD8⁺ T cells isolated from PBMC obtained from HLA-A2⁺ normal donors using peptide-pulsed autologous dendritic cells as the APC [20]. IVS of CD8⁺ T cells obtained from 2 of 3 HLA-A2⁺ donors yielded effector cells that essentially did not significantly recognize HSD17B12 _114–122 peptide-pulsed T2 cells compared with irrelevant HIV peptide-pulsed or unpulsed T2 target cells in ELISPOT IFNγ assays (Fig. 1). Further stimulation of these effector cells using peptide-pulsed PBMC as APC repeatedly resulted in their loss of activity, and attempts by single cell dilution to obtain enriched stable populations of HSD17B12_114–122 peptide-specific CD8⁺ T cells from these IVS populations failed (data not shown). Therefore, using a technique we and other have used to enhance the immunogenicity of an epitope, we investigated the possibility that the Meth A Hsd17b12^114T peptide (TYDKIKTGL) might act as an optimized peptide in inducing CD8⁺ T cells cross-reactive against the HSD17B12_114–122- peptide [21–23]. IVS of CD8⁺T cells isolated from PBMC obtained from 3 of 4 HLA-A2⁺ normal donors using autologous dendritic cells pulsed with the Hsd17b12^114T peptide yielded effector cells that recognized the Hsd17b12^114T peptide-pulsed T2 target cells, but most importantly, also recognized HSD17B12_114–122 peptide-pulsed T2 cells compared with irrelevant HIV peptide-pulsed or unpulsed T2 target cells in ELISPOT IFNγ assays (Fig. 1).

Recognition of naturally presented HSD17B12_114–122peptide by HLA-A2-restricted CD8⁺T cells cross-reactive against the HSD17B12_114–122peptide. The ability of HLA-A2-restricted CD8⁺ T cells reactive against HSD17B12_114–122 peptide to recognized HLA-A2⁺ human carcinoma cells naturally presenting the HSD17B12_114–122 peptide was tested using panels of human SCCHN, breast and ovarian carcinoma cell lines as target cells in ELISPOT IFNγ assays. All the cell lines were pretreated with IFNγ to enhance their presentation of HLA class I antigen/tumor peptide complexes for recognition by the CD8⁺ T cells [24]. Five SCCHN cell lines were tested: the HLA-A2⁺ PCI-13, SCC-4, UD-SCC6, and PCI-30 and HLA-A2^neg PCI-4A (Fig. 2a). Of these five SCCHN cell lines, the HLA-A2⁺ PCI-13, SCC-4, UD-SCC6 were recognized, while PCI-30 and HLA-A2^neg PCI-4A were not (Fig. 2b). The recognition of the HLA-A2⁺ PCI-13, SCC-4, UD-SCC6 cell lines was blocked by the HLA A2,-A28-specific mAb KS1, but not the HLA-DR-spe

cific mAb L243, confirming the HLA-A2 restriction of the effectors. The epitope specificity of the effector cells used for these experiments was indicated by their recognition of HSD17B12_114–122 peptide-pulsed T2 target cells but not the irrelevant HIV-pulsed T2 target cells (Fig. 2b). Three HLA-A2⁺ breast carcinoma cell lines, MDA-MB-231, MDA-MB-453 and MCF-7 cell lines, and three HLA-A2⁺ ovarian carcinoma cell lines, AD10, OVCAR429 and OVCAR3, were also tested for recognition by these effector cells. The CD8⁺ T cells recognized all three breast carcinoma cell lines in ELISPOT INFγ assays (Fig. 2b). The recognition of the HLA-A2⁺ breast carcinoma cell lines was blocked by the HLA A2,-A28-specific mAb KS1, but not the HLA-DR-specific mAb L243. In contrast to these findings, the effector cells recognized none of the ovarian carcinoma cell lines (data not shown). Further analysis indicated that HLA class I expression was not upregulated in the ovarian carcinoma cell lines by IFNγ -pretreatment (data not shown). This finding is consistent with reports of severe defects in antigen processing and presentation in ovarian carcinomas [25–27].

Critical to the potential use of “self” tumor antigens in cancer vaccines, the HSD17B12-reactive CD8⁺ T cells did not recognize JL-BRL 6 cells, a low-passage cell line established from the HLA-A2⁺ normal breast reduction tissue, even when pretreated with IFNγ, which upregulates HLA class I antigen expression [24] (Fig. 2b). In addition, the CD8⁺ T cell effectors did not recognize HLA-A2⁺ human lung fibroblast cells, MRC-5, unless the cells were transfected with HSD17B12 cDNA (Fig. 2c). Reactivity of the effector cells against the transfected fibroblasts was blocked by the HLA-A2,-A28-specific mAb KS1, but not the HLA-DR-specific mAb L243. Together with the results obtained using normal human fibroblasts, the results strongly support the ability of HSD17B12-specific CD8⁺ T cells to recognize most types of tumor cells overexpressing HSD17B12, but not normal cells.

Analysis of HSD17B12 expression in human carcinoma cell lines and normal cells

The expression of HSD17B12 at the protein level in cell-free extracts of the human SCCHN, breast, and ovarian carcinoma cell lines, as well as the JL-BRL 6 cell line used in this study was evaluated by immunoblot analysis using the rabbit anti-TYDKIKTGL antibody (Fig. 3). Based on a sequence of 312 amino acids, HSD17B12 should be detected as M_r35 kDa species in immunoblot assays. In most of tumor cell-derived extracts analyzed, however, a doublet consisting of a M_r 35 kDa and a M_r 40 kDa species was detected. The higher molecular weight species was prominent in the most of the breast, ovarian, and SCCHN cell lines analyzed. In the case of the SCCHN SCC-4 cell, the higher M_r species, itself, appeared to be a doublet. The exceptions were the UD-SCC 6 and OVCAR-3 cell lines; the latter expressing the lowest level of HSD17B12 of any of the tumor cell lines analyzed. Only the lower M_r HSD17B12 species was detected in the cell-free extract of the non-transformed, low-passage breast reduction JL-BRL 6 cell line. The significance of these findings requires further study, but they suggest that processing and post-translational modifications of HSD17B12, namely glycosylation, are probably different in the tumor cells compared with the non-transformed “normal” cells studied. Based on the results of this immunoblot analysis, with the exception of the UD-SCC 6 cell line, recognition of a tumor cells overexpressing HSD17B12 correlated with expression of high levels of the higher rather than lower M _r HSD17B12 species (Figs. 2 and 3).

Immunohistochemistry analysis of HSD17B12 expression in SCCHN

HSD17B12 expression in an SCCHN cell line and two SCCHN TMA was analyzed by immunohistochemistry using the rabbit anti-TYDKIKTGL antiserum. The specificity of the antiserum for HSD17B12 was established using paraffin-embedded sections of formalin-fixed PCI-13 cells (Fig. 4). It showed predominate cytoplasmic staining with some nuclear staining of PCI-13 cells, which were derived from a poorly differentiated SCCHN lesion [17]. Comparable staining was also obtained with an SCCHN lesion. In the presence of a fixed concentration of blocking peptide, staining of PCI-13 was completely blocked and significant reduced in the lesion; indicative of the specificity of this antiserum. The first SCCHN TMA analyzed consisted exclusively of 15 oral cavity lesions with a range of tumor grades; nine well-differentiated, four moderately differentiated, and two poorly differentiated, all with adjacent normal mucosa tissue. HSD17B12 expression was detected in all the lesions, whereas the non-neoplastic squamous mucosa showed absence of expression (Fig. 5a). HSD17B12 expression was detected in the cytoplasm of all the lesions in this TMA. All of the well-differentiated carcinomas also showed some degree of nuclear localization (Fig. 5b). Of the four moderately differentiated SCCHN, two had nuclear/cytoplasmic staining (Fig. 5c), the other two had cytoplasmic expression only. Of the two poorly differentiated SCCHN, one showed cytoplasmic staining, the other cytoplasmic/nuclear staining (Fig. 5d).

The IHC analysis of a second SCCHN TMA, a commercially available TMA consisting of specimens of 48 lesions of various sites, only 25% of which were oral cavity tumors. The clinicopathological characteristics of the specimens in this TMA, which consisted of sex and age of the subjects, tumor site, grade, and stage, as well as nodal involvement and metastasis, are listed in Table 1. Varied levels of HSD17B12 expression were detected in 44/48 (91.7%) specimens in the TMA. Moderate staining was detected in 13/44 (29.5%) of specimens and strong staining in 19/44 (43.2%). As noted with the previous oral cavity TMA, the staining of the multisite TMA also showed varied cellular localization of HSD17B12 expression among the specimens. A mixed cellular localization was evident in 34/44 specimens (77.3%). HSD17B12 expression in laryngeal tumors was significant compared with other sites, and in poorly differentiated tumors compared with moderate- and well-differentiated lesions. The results of staining of both TMA, therefore, are consistent with HSD17B12 overexpression being (1) associated with neoplasia, (2) present in nucleus and cytoplasm, and (3) predominate detected in poorly differentiated SCCHN lesions. These findings regarding nuclear as well as cytoplasmic cellular localization of HSD17B12 expression in SCCHN cell lines and lesions are in contrast to those of other laboratories involving breast carcinoma cell lines and lesions [11, 13, 14] and warrant further consideration.

Multifunctional role of HSD17B12 in PCI-13 cells

The role of HSD17B12 activity in E1 to E2 conversion and elongation of long-chain fatty acids in PCI-13 cells was investigated using the approach detailed by Nagasaki et al. [14]. In a series of experiments, we determined that inhibition of HSD17B12 mRNA in PCI-13 cells required transfection of the minimum number of cells needed for subsequent analysis at the maximum recommended, non-toxic concentration of 8 μg/ml of HSD17B12 siRNA. At this concentration, a 50% decrease in HSD17B12 mRNA was associated with a 38% decrease in cell growth (Fig. 6). The inhibition of cell growth was reversed by either AA or E2. Presumably, if functional in both pathways, one would have expected that both agents be required to reverse the growth inhibition due to HSD17B12 knockdown. Since only a maximum of a 50% decrease in HSD17B12 mRNA could be achieved using this approach; however, it is plausible that there remained sufficient residual HSD17B12 activity not to require this. Although the participation of HSD17B12 in these pathways was not directly studied, the result does support HSD17B12 functional activity in E1/E2 conversion as well as fatty acid elongation in the PCI-13 cell line and, presumably, SCCHN.